Glucose measurement in the intensive care unit

Fingerstick (capillary blood) samples are not suitable for glucose measurement in an intensive care setting and whole-blood glucose results obtained using a blood gas analyzer agree more closely with reference laboratory measurement than a dedicated point-of-care glucose meter. These are the two headline conclusions of the most recently published study of point-of-care glucose testing. 

The study population comprised 84 patients being cared for at an intensive care unit where patients are subject to a "tight glycemic control" protocol. Venous, arterial and capillary blood were sampled simultaneously from each study patient (a total of 114 sample sets). 

Venous and arterial blood were submitted for glucose measurement by three testing modalities: a dedicated point-of-care handheld glucose meter (whole-blood measurement), a blood gas analyzer (whole-blood measurement) and a central laboratory analyzer (plasma measurement). The capillary blood samples were submitted for glucose meter measurement only. 

The mean glucose levels of arterial/venous samples using the blood gas analyzer showed a bias of 0.1-0.3 mmol/L compared with the reference laboratory method. The bias was greater (0.7-0.9 mmol/L) for arterial/venous and capillary samples measured using the glucose meter. 

Further statistical analysis (Parkes error grid) revealed that there was no clinically significant difference between plasma results obtained by the reference method and those obtained from blood gas analyzer (for both arterial and venous blood samples). The same could be said of the glucose meter results for both arterial and venous blood. 

However, there was a clinically significant difference for results when capillary blood glucose measured on the glucose meter was compared with the reference method. In three cases capillary blood analysis severely underestimated the glucose concentration (<46 % of laboratory result). The authors conclude that they cannot recommend the use of capillary (fingerstick) samples to monitor glucose levels among intensive care patients.